Blog
Insights on food delivery search, menu intelligence, and building AI for restaurant platforms.
Your Embedding Model Thinks Bread and Honeycomb Are the Same Food
General-purpose embeddings confuse co-occurrence with identity. Bread and honeycomb share a breakfast table but belong to completely different food categories. Here's why that breaks search.
Introducing FoodEval: a benchmark for food domain embeddings
MTEB and BEIR contain zero food evaluations, so a model can top the leaderboard and still miss that Paneer Tikka and Cottage Cheese Tikka are the same dish. FoodEval measures the food gap. It is public.
Production System Benchmarks
How Latimal's production search and matching pipeline compares to leading embedding models on real food-domain tasks: search, matching, and classification.
Why Keyword Search Fails for Food Delivery
When a customer types 'cold coffee', keyword search can't find 'Iced Americano'. Food delivery has a discovery problem that string matching will never solve.
Building Semantic Search for Restaurant Menus
A practical guide to replacing keyword search with semantic search in food delivery apps. Two integration paths, pre-computed embeddings, and real code.
The Hidden Cost of Duplicate Menu Listings
A single dish can appear under five different names across POS systems, aggregators, and languages. The business cost is bigger than most platforms realize.
Smart Cart Upsell: Beyond 'Customers Also Bought'
Collaborative filtering doesn't work for food. Context-aware pairing does. How to build upsell that understands what goes with what.